Machine translation has become an integral part of online life in recent years. The different systems have grown significantly, allowing them to offer increasingly better results. The quality of machine translation has also improved significantly, primarily due to advances in artificial intelligence and machine learning.
What is machine translation?
Machine translation (MT) is the automatic translation of text, based on a combination of computer systems and artificial intelligence. The most modern and widely used approach is neural network translation (NMT), based around deep neural networks. NMT models learn how to translate specific texts by analysing large amounts of training data, allowing them to better understand both the context and the language structure.
In addition, there is an older approach to machine translation, known as statistical machine translation (SMT). This method relies on the analysis of large linguistic data sets to detect translation regularities. Statistical models are then used to learn what words or phrases often occur together in different languages, and then make translations based on this.
Why do we opt for website machine translation?
Websites are often a user’s first contact with a brand or content. The translation of the website into various languages has since become an indispensable part of a company’s expansion into the international market. One of the ways in which this process is significantly facilitated is through machine translation.
Standard website translation can take a long time and be expensive. In order to speed up the entire translation process and reduce the costs, many companies are opting for an NMT solution.
If the stylistic correctness of the text and fluency are not the most important, but the key is to understand the source text, then NMT is the ideal solution. Machine translation can be particularly beneficial for longer, repetitive projects or extensive websites such as online shops, as well as for projects where the task is translating user manuals.
Machine translation of web pages
Here are some of the steps showing how to achieve a professional machine translation of a website.
1. Choosing the right translation platform
The first step to creating a professional machine translation of a website is to choose the right translation platform. There are many tools available, such as Google Translate, DeepL, Microsoft Translator and many others.
2. Choosing the target languages
The next step is to determine into which languages we want to translate the website.
3. Preparing the content for translation
It is very a good idea to prepare the content of each page carefully before starting the translation process.
4. Using the available translation tools
Once we have chosen a translation platform and prepared the content, we can proceed with the actual translation process. Machine translation tools are easy to use – just paste in the source text, select the target language and you have your translation. However, it is important to bear in mind that, although the machine translation process is always improving, there may still be some errors and inaccuracies.
Popular CMS systems that are used to create websites often provide auto-translation plug-ins that allow the entire page to be translated almost instantly, preserving its layout, without the need to copy the content into a translation tool.
5. Correction and adjustment
Machine translations always require some form of proofreading and adjustment. Reviewing the translation is essential to ensure that it is understandable and corresponds to the tone of the original text, correcting any errors and checking for any inaccuracies along the way. It is also possible to use post-editing of machine translations to achieve professional proofreading of a machine translation.
6. Testing and optimisation
Once the website has been translated, it is worth testing it for design and functionality. We make sure that all elements of the site are correctly displayed and that the links and navigation system work correctly. It is also worth optimising the content for SEO to ensure the visibility of the site in the search results.
Post-delivery of machine translations
Post-translation of machine translations is similar to proofreading of the machine translation by a professional translator. The main aim of post-editing is to improve the quality of the translation so that it is more precise, understandable and appropriately worded in the target language.
It is often argued that machine translation will replace human labour, that the translation profession will not soon be required. Can we therefore see a threat to the work of translators in the development of new technologies and neural machine translation? Post-delivery machine translation can be viewed as an important step in the process of producing high-quality translations. While machine translation can significantly speed up the translation process, professional translators and editors still play an important role in improving the translations to make them more accurate and tailored to the audience. Machine translation is therefore no substitute for the work done by humans.
Types of post-editing for machine translation
There are two types of post-editing of machine translations: light post-editing and full post-editing. Light post-editing aims to produce a text of a quality that allows the text to be understood and basic linguistic and grammatical errors to be eliminated, while full post-editing aims to rewrite the text so that its quality is close to that of a human translated text. The post-editor often has to rephrase the content so that it conveys the character or emotional message of the source text. This is particularly desirable for texts intended for publication.
Is it worth choosing machine translation?
Using machine translation to translate an entire website can be beneficial in some cases. It is certainly worth considering a number of factors at the outset before deciding on the machine translation of a website.
Translation quality: Machine translation does not always provide high-quality translations, especially if the website contains complex or specialised content.
Cultural context: If the site contains content related to a particular culture or region, then machine translation may overlook the linguistic and cultural subtleties. In this case, post-editing by the translator or localisation of the translation may be necessary.
Security and privacy: If a website contains sensitive information or the personal data of the users, it is necessary to consider security and privacy issues when using online machine translation tools.
Technical support: If there are problems with the machine translation of the website, it is essential to have access to appropriate technical support to resolve any problems.
Monitoring and updates: Machine translation is not a static process. It is worth monitoring the translations on the website regularly and updating them as necessary to ensure ongoing quality.
Alternatives: In some cases, especially if the budget allows, it may be better to hire a professional translator or translation agency that offers translation services of high quality and accuracy.
In conclusion, it is worth using machine translation on a website if you are aware of its limitations and potential problems. However, for specialised content or content requiring precision, it is worth investing in a professional translation to ensure the best quality and effectiveness of communication with the website users.
Integrating CAT tools with NMT
It should also be mentioned that many of the tools that interpreters use on a daily basis, known as CAT tools (computer-aided translation), offer the possibility of integrating with NMT engines such as Google Translate, Deepl, etc. When working in CAT tools, it is possible to combine multiple options like translation memories and the NMT engine, which saves time and reduces the cost of the translation, as well as maintaining text consistency.
The conclusion is that professional machine translation of a website is possible, but requires proper preparation, proofreading and monitoring.
Machine translation at Skrivanek
Skrivanek’s extensive range of services includes human and machine translations, no matter what the source language, from one language to another. We offer raw neural machine translation, proofreading and post-editing in different variants, with a range of cat tool options.
As one of the major language service providers, we use the machine translation engine that we have developed – Skrivanek NMT. We also offer computer assisted translation, which in combination with translation memories, can significantly reduce the price. Our language tools not only help us to maintain the consistency and meaning of the text, but also to provide the opportunity to offer discounts for translations featuring repetitive content. We offer website translations using nmt engines, as well as post edit testing of the finished translated website.
The machine translation of a website can be problematic, especially for non-IT professionals. It can also be a source of technical problems, including translation gaps and content import problems. At Skrivanek we can ensure that you receive a correct translation by the way we carry out the entire process from the technical perspective.
Write to us so we can prepare a quotation for translating your website.